On 20.10.2016 09:19, Raffael Sahli wrote:

Hi

What can lead a consumer to "randomly" delete ~50% of all objects in his database? I have this problem now for ~1-2months and on 3 different master/slave groups.

The consumer starts to delete objects (but those are all present on the master):

Oct 19 17:16:22 ldap-slave002.xxx slapd[8554]: do_syncrep2: rid=999 LDAP_RES_INTERMEDIATE - SYNC_ID_SET Oct 19 17:16:22 ldap-slave002.xxx slapd[8554]: do_syncrep2: rid=999 LDAP_RES_INTERMEDIATE - SYNC_ID_SET Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID 9c802b18-7c2c-1033-9f25-2d4a65066d19, dn uid=jasmin.mans,ou=none,o=dd,dc=xxx,dc=xxx Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID 650ada00-e996-1035-80ac-a173474274df, dn uid=jasmine.suddr,ou=none,o=dd,dc=xxx,dc=xxx Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID ec5c9aa4-7c2c-1033-9fa4-2d4a65066d19, dn uid=jonas.beffrnn,ou=none,o=dd,dc=xxx,dc=xxx Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID 7461d22e-e996-1035-80f6-a173474274df, dn uid=juerg.klegegeli,ou=none,o=dd,dc=xxx,dc=xxx Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID 758c744c-e996-1035-8100-a173474274df, dn uid=julia.sxgegegea,ou=none,o=dd,dc=xxx,dc=xxx Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID 47e4f6b4-7c2d-1033-8015-2d4a65066d19, dn uid=karin.vivwvwe,ou=none,o=dd,dc=xxx,dc=xxx Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID 8d26a4f2-7c2d-1033-807b-2d4a65066d19, dn uid=larissa.fewfewf,ou=none,o=dd,dc=xxx,dc=xxx Oct 19 17:16:24 ldap-slave002.xxx slapd[8554]: nonpresent_callback: rid=999 nonpresent UUID 98310aa4-7c2d-1033-808b-2d4a65066d19, dn uid=laura.cfwfew,ou=none,o=dd,dc=xxx,dc=xxx

syncrepl config: olcSyncrepl: {0}rid=999 provider=ldap://ldap-master002.xxx.xxx:389 bindmethod=simple timeout=0 network-timeout=10 binddn="uid=replicator,ou=system,o=de,dc=xxx,dc=xxx" credentials="pass" keepalive=0:0:0 starttls=yes tls_cacert="/etc/ssl/certs/SwissSign_Silver_CA_-_G2.pem" tls_reqcert=allow filter="(objectClass=*)" searchbase="dc=xxx,dc=xxx" scope=sub attrs="*,+" syncdata="accesslog" logbase="cn=accesslog_xxx" logfilter="(&(objectClass=auditWriteObject)(reqResult=0))" schemachecking=off type=refreshOnly interval="00:00:01:00" retry="10 5 60 +"

Misconfiguration? (We've been running this configuration for years...)

Workaround is to remove the whole mdb database and restart slapd to let it resync. I've found some posts with a similar problem but those are all related to pre 2.4.21, we're running slapd 2.4.44.

Cheers

Ideas? (Most of the time slapd removes now ~99% of all objects)